Deep links can open the right screen in a mobile app, but campaign tracking becomes harder when some people have the app, others do not, and some arrive from desktop. This guide shows how to design trackable mobile app campaign links, estimate the outcomes you can measure, choose reliable fallbacks, and run a repeatable release QA process before links reach customers.
Overview
A mobile campaign link has two jobs: route a person to the most useful destination and preserve enough campaign context for attribution. If the app is installed, that destination may be a product page, article, checkout screen, invitation, or account area inside the app. If the app is not installed, the link may need to send the person to an app store, a mobile website, or a campaign-specific landing page. Desktop visitors usually need a separate web destination.
These routing rules are often called deep linking. Universal Links are Apple-platform links that can open an installed app through an associated website domain. Android App Links provide a comparable app-opening path on Android. Deferred deep linking attempts to preserve the intended destination through installation, so a new user can reach the relevant in-app screen after opening the app for the first time. A standard web redirect, by contrast, sends the visitor to a URL without automatically providing app-specific behavior.
| Approach | Installed app | App not installed | Best use |
|---|---|---|---|
| Universal Links | Opens the associated iOS app when configured correctly | Falls back to the website | iOS campaigns with a controlled web fallback |
| Android App Links | Opens the associated Android app when verified | Falls back to the web destination | Android campaigns with app-aware routing |
| Deferred deep links | Opens the intended in-app location | Can preserve campaign intent through installation, depending on the implementation | App-install campaigns and new-user onboarding |
| Standard web redirect | May open a browser rather than the app | Opens the specified webpage | Simple campaigns, fallback links, or environments without app-link support |
The exact behavior depends on implementation, operating-system settings, browser behavior, app state, and analytics configuration. Treat each route as something to test, not something to assume.
How to estimate
Use a simple funnel model to estimate what a campaign link can tell you. Start with link clicks, then separate the routes taken by app-installed users, non-installed users, and desktop visitors.
Estimated app opens = total valid clicks × installed-app rate × app-open rate
Estimated store visits = total valid clicks × non-installed rate × store-fallback rate
Estimated web sessions = total valid clicks × web-fallback rate
For campaign decisions, add conversion rates:
Estimated conversions = route-specific visitors × route-specific conversion rate
Use measured historical values where they exist. If they do not, enter low, expected, and high assumptions rather than presenting one forecast as a fact. Keep the calculation by route because an in-app purchase flow, an app-store visit, and a web landing page are different experiences.
Do not confuse a click with an app open, installation, or conversion. A short link or redirect can count a request, while an analytics SDK may record an app open only after the app initializes. Privacy controls, blocked scripts, duplicate clicks, and delayed events can also create differences between systems. Use the model to plan measurement and compare scenarios, not to promise a result.
Inputs and assumptions
Record the following inputs in a campaign sheet before creating the link:
- Campaign identity: campaign name, launch date, creative, placement, audience, and owner.
- Source and medium: for example, newsletter, paid social, creator post, QR code, or in-app message. Use one consistent naming convention.
- Content identifier: distinguish each ad, email button, influencer, placement, or creative variation.
- Primary destination: the exact in-app path, such as
/offers/springor a product identifier. - iOS fallback: choose the relevant webpage, App Store destination, or campaign landing page.
- Android fallback: choose the relevant webpage, Play Store destination, or campaign landing page.
- Desktop destination: specify a useful web page rather than relying on a generic homepage.
- UTM parameters: define
utm_source,utm_medium,utm_campaign, and, where useful,utm_contentorutm_term. - Analytics events: link click, app open, first open, install or activation where available, destination view, sign-up, purchase, or another agreed conversion.
- Attribution window: document the period used to connect a click with a later action. The window should match your analytics setup and reporting purpose.
A reusable mapping template can look like this:
campaign: spring_offer source: newsletter medium: email content: hero_button iOS_app_path: app://offers/spring iOS_fallback: https://example.com/spring Android_app_path: app://offers/spring Android_fallback: https://example.com/spring desktop_destination: https://example.com/spring utm_campaign: spring_offer utm_source: newsletter utm_medium: email utm_content: hero_button
Keep sensitive information out of query strings. Use opaque campaign or user identifiers only when your privacy, security, and analytics teams have approved the design. A privacy-first link-tracking workflow should collect only what is needed, respect applicable consent choices, and avoid treating inferred device or identity data as perfectly complete.
Worked examples
Example 1: An app-installed campaign
Suppose a campaign generates 10,000 valid clicks. For planning purposes, your team assumes that 60% come from people with the app installed, 90% of those users successfully open the app, and 25% of app openers complete the target action.
The calculation is:
- Installed-app clicks: 10,000 × 60% = 6,000.
- Estimated app opens: 6,000 × 90% = 5,400.
- Estimated conversions: 5,400 × 25% = 1,350.
These are assumptions, not reported results. After launch, replace them with observed values and investigate gaps between click, app open, and conversion events.
Example 2: An app-install campaign with fallbacks
Now assume 10,000 valid clicks, with 40% from people who do not have the app installed. If 70% of that group reaches the store destination and 20% of store visitors install and activate, the planning estimate is:
- Non-installed clicks: 10,000 × 40% = 4,000.
- Estimated store visits: 4,000 × 70% = 2,800.
- Estimated activations: 2,800 × 20% = 560.
That estimate does not prove that the link caused every activation. Store reporting, app analytics, consent settings, and attribution rules may produce different totals. Report the assumptions beside the result so another person can reproduce or revise the calculation.
For naming and campaign consistency, compare this setup with the UTM parameters and campaign tracking template. If the link will be distributed through email, the email tracking and QA checklist provides a useful parallel workflow.
When to recalculate
Recalculate the model and repeat testing whenever an input changes. That includes a new app release, a changed app route, a new store listing, a redesigned fallback page, a new attribution rule, or a change in campaign audience or creative. Revisit it when click volume, app-open rates, store visits, activation rates, or conversion rates move enough to affect a decision. Pricing changes for analytics or link-management tools can also change the cost side of a campaign evaluation, so record the tool and plan assumptions used in each review.
Before every launch, run this release QA sequence:
- Create a unique trackable link and confirm its UTM values and destination mapping.
- Test on iOS with the app installed, then with the app unavailable.
- Test on Android with the app installed, then with the app unavailable.
- Test a desktop browser and verify that the desktop destination is intentional.
- Test a fresh install or first-open path if deferred deep linking is part of the campaign.
- Check that the intended in-app screen opens, not merely the app home screen.
- Confirm that click, open, destination, and conversion events appear with the expected campaign values.
- Repeat tests from email, social, QR code, and any other real distribution context because wrappers and browsers can alter behavior.
- Record device, operating-system version, app version, browser, result, and defect owner.
- After launch, compare redirect logs, link analytics, app analytics, store reporting, and conversion data without assuming that totals will match exactly.
Store the final mapping and test evidence with the campaign record. For ongoing maintenance, use the short link audit checklist to find broken redirects and stale destinations. A deep-link system becomes dependable when its assumptions, fallbacks, and measurement gaps are documented and recalculated as the app and campaign change.